Last week, we looked at Mariana Minerals, a company attempting to rebuild U.S. critical-minerals production under severe time, labor, and political constraints.

This week: The Career Risk That Kills Defense Innovation

As part of the 2025 Defense Tech Signals Coverage Map, I pulled total historical DoD obligations for 20 venture-backed defense startups. This isn’t an exhaustive list but it’s representative of the sector.

Their total DoD obligations: $5.5 billion. Anduril Industries alone accounts for $4.3 billion of that. Let’s be generous. Double the number to account for the rest of the ecosystem and call it $8.5 billion.

Last year’s defense budget alone was $850 billion. In the context of defense contracts, $8.5B in obligations is a rounding error.

The usual explanation to this is bureaucracy. The FAR is too complex. Procurement timelines are glacial. Compliance requirements are crushing. If only we could reform these systems, innovation would finally flow.

That explanation is convenient. And wrong.

You can hire people to navigate FAR compliance extremely well. Entire careers exist to do exactly that, and if you pay for the expertise, it’s available. And reforms have been made that opened doors for smaller players who can’t afford elite business-development teams.

And it ultimately masks a real problem:

Rational risk aversion in program offices, and the asymmetric career consequences of "failure".

When a program manager awards a contract to one of the Big Five and the program runs late and over budget, that's just defense contracting. Schedule extensions happen. Cost overruns get negotiated. Heads don't roll because this is how it's always worked.

Now consider the alternative. A program manager bets on a non-traditional, and the program fails.

“What do you mean you spent taxpayer dollars on an unproven platform?
You degraded our ability to deter aggression.
Don't be a tall poppy”

That can be a career-ending decision.

Yes, things are changing. Neo-primes and non-traditionals are winning OTAs and CSOs and CRADAs. But we've yet to see someone take home a major prime contract, the big programs that actually move the needle on defense spending.

Take the Collaborative Combat Aircraft program.

Nobody, not Anduril, not General Atomics, not Boeing, not Northrop, not Kratos, has built an affordable, autonomous wingman operating at scale in contested airspace with advanced IADS. This is genuinely new for everyone.

But when contracts get awarded, the program office will face a choice: Do you bet on a company with a few billion dollars in total DoD obligations or on a firm with 80 years of building carrier-capable aircraft.

Both are exceptionally unproven for this mission.  But only one choice gets you fired if it fails. Is that wrong?

If you're the program manager and Anduril can't deliver CCAs at scale, you've blown the budget on platforms that can't survive in theater. If Boeing can't deliver, well, Boeing couldn't deliver. There's a difference.

The upside likely isn’t very high either. If the startup succeeds, you get a bullet on a performance report. If the prime succeeds, that’s just baseline performance.

The deeper problem is that risk gets diffused through so many bureaucratic layers that nobody has to own the decision.

-The Contracting Officer points up: "I executed the acquisition strategy approved by the Program Manager."
-The Program Manager points up: "I executed validated requirements."
-The Requirements Officer points up: "I translated operational guidance."
-The Operational Commander points up: "I prioritized based on strategic direction."

Everyone followed process. Nobody made a call. When the process reliably leads to the safest choice, that's not bureaucracy failing. It's actually bureaucracy working exactly as designed.

You want different outcomes? You have to change incentives at every node. Right now, there is no upside to taking risk and massive downside to being wrong.

As officers and civilians move up in rank and seniority, risk tolerance declines. That's human. Family obligations, financial stability, and time horizons matter. But the institution itself enables this.

We need a culture shift to: "Nobody got fired for choosing the solution they genuinely believed would deliver better capability, even if it was unproven."

This isn't actually an argument that Primes don't deliver because they do. Just slowly and expensively. The F/A-18 works. The F-35 works. Virginia-class subs work. "Works eventually at high cost" is better than "might not work at any cost."

I'm also not here to make the argument that non-traditionals are obviously better. Most are unproven at the scale the Department of Defense requires. Venture funding and prototype demos are not the same as full-rate production under operational constraints.

And ultimately, this isn't an argument that bureaucracy doesn't matter. It does. But it's not the binding constraint everyone says it is. You can hire people to navigate FAR. What you can't navigate is a system where choosing a startup over an incumbent can end a career.

We've spent years blaming acquisition regulations for why defense startups can't break through. We've reformed OTAs, stood up innovation units, and created fast-track procurement vehicles.

And startups still have less than 1% market share.

The problem isn't that program offices can't choose non-traditionals. It's that they won't, because failure is punished asymmetrically. When a startup fails, you get fired. When a prime fails, you get a schedule extension.

Until that incentive structure changes with real protection for judgment, real accountability for outcomes, and real operator input, the money will keep flowing to the safest option.

CCA will tell us whether anything has actually changed, or whether we've just created new letterhead for the same old game.